1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * Copyright (C) 2007 Oracle. All rights reserved. 4 */ 5 6 #include <linux/bio.h> 7 #include <linux/slab.h> 8 #include <linux/pagemap.h> 9 #include <linux/highmem.h> 10 #include "ctree.h" 11 #include "disk-io.h" 12 #include "transaction.h" 13 #include "volumes.h" 14 #include "print-tree.h" 15 #include "compression.h" 16 17 #define __MAX_CSUM_ITEMS(r, size) ((unsigned long)(((BTRFS_LEAF_DATA_SIZE(r) - \ 18 sizeof(struct btrfs_item) * 2) / \ 19 size) - 1)) 20 21 #define MAX_CSUM_ITEMS(r, size) (min_t(u32, __MAX_CSUM_ITEMS(r, size), \ 22 PAGE_SIZE)) 23 24 #define MAX_ORDERED_SUM_BYTES(fs_info) ((PAGE_SIZE - \ 25 sizeof(struct btrfs_ordered_sum)) / \ 26 sizeof(u32) * (fs_info)->sectorsize) 27 28 int btrfs_insert_file_extent(struct btrfs_trans_handle *trans, 29 struct btrfs_root *root, 30 u64 objectid, u64 pos, 31 u64 disk_offset, u64 disk_num_bytes, 32 u64 num_bytes, u64 offset, u64 ram_bytes, 33 u8 compression, u8 encryption, u16 other_encoding) 34 { 35 int ret = 0; 36 struct btrfs_file_extent_item *item; 37 struct btrfs_key file_key; 38 struct btrfs_path *path; 39 struct extent_buffer *leaf; 40 41 path = btrfs_alloc_path(); 42 if (!path) 43 return -ENOMEM; 44 file_key.objectid = objectid; 45 file_key.offset = pos; 46 file_key.type = BTRFS_EXTENT_DATA_KEY; 47 48 path->leave_spinning = 1; 49 ret = btrfs_insert_empty_item(trans, root, path, &file_key, 50 sizeof(*item)); 51 if (ret < 0) 52 goto out; 53 BUG_ON(ret); /* Can't happen */ 54 leaf = path->nodes[0]; 55 item = btrfs_item_ptr(leaf, path->slots[0], 56 struct btrfs_file_extent_item); 57 btrfs_set_file_extent_disk_bytenr(leaf, item, disk_offset); 58 btrfs_set_file_extent_disk_num_bytes(leaf, item, disk_num_bytes); 59 btrfs_set_file_extent_offset(leaf, item, offset); 60 btrfs_set_file_extent_num_bytes(leaf, item, num_bytes); 61 btrfs_set_file_extent_ram_bytes(leaf, item, ram_bytes); 62 btrfs_set_file_extent_generation(leaf, item, trans->transid); 63 btrfs_set_file_extent_type(leaf, item, BTRFS_FILE_EXTENT_REG); 64 btrfs_set_file_extent_compression(leaf, item, compression); 65 btrfs_set_file_extent_encryption(leaf, item, encryption); 66 btrfs_set_file_extent_other_encoding(leaf, item, other_encoding); 67 68 btrfs_mark_buffer_dirty(leaf); 69 out: 70 btrfs_free_path(path); 71 return ret; 72 } 73 74 static struct btrfs_csum_item * 75 btrfs_lookup_csum(struct btrfs_trans_handle *trans, 76 struct btrfs_root *root, 77 struct btrfs_path *path, 78 u64 bytenr, int cow) 79 { 80 struct btrfs_fs_info *fs_info = root->fs_info; 81 int ret; 82 struct btrfs_key file_key; 83 struct btrfs_key found_key; 84 struct btrfs_csum_item *item; 85 struct extent_buffer *leaf; 86 u64 csum_offset = 0; 87 u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); 88 int csums_in_item; 89 90 file_key.objectid = BTRFS_EXTENT_CSUM_OBJECTID; 91 file_key.offset = bytenr; 92 file_key.type = BTRFS_EXTENT_CSUM_KEY; 93 ret = btrfs_search_slot(trans, root, &file_key, path, 0, cow); 94 if (ret < 0) 95 goto fail; 96 leaf = path->nodes[0]; 97 if (ret > 0) { 98 ret = 1; 99 if (path->slots[0] == 0) 100 goto fail; 101 path->slots[0]--; 102 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); 103 if (found_key.type != BTRFS_EXTENT_CSUM_KEY) 104 goto fail; 105 106 csum_offset = (bytenr - found_key.offset) >> 107 fs_info->sb->s_blocksize_bits; 108 csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]); 109 csums_in_item /= csum_size; 110 111 if (csum_offset == csums_in_item) { 112 ret = -EFBIG; 113 goto fail; 114 } else if (csum_offset > csums_in_item) { 115 goto fail; 116 } 117 } 118 item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item); 119 item = (struct btrfs_csum_item *)((unsigned char *)item + 120 csum_offset * csum_size); 121 return item; 122 fail: 123 if (ret > 0) 124 ret = -ENOENT; 125 return ERR_PTR(ret); 126 } 127 128 int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans, 129 struct btrfs_root *root, 130 struct btrfs_path *path, u64 objectid, 131 u64 offset, int mod) 132 { 133 int ret; 134 struct btrfs_key file_key; 135 int ins_len = mod < 0 ? -1 : 0; 136 int cow = mod != 0; 137 138 file_key.objectid = objectid; 139 file_key.offset = offset; 140 file_key.type = BTRFS_EXTENT_DATA_KEY; 141 ret = btrfs_search_slot(trans, root, &file_key, path, ins_len, cow); 142 return ret; 143 } 144 145 static blk_status_t __btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio, 146 u64 logical_offset, u32 *dst, int dio) 147 { 148 struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); 149 struct bio_vec bvec; 150 struct bvec_iter iter; 151 struct btrfs_io_bio *btrfs_bio = btrfs_io_bio(bio); 152 struct btrfs_csum_item *item = NULL; 153 struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; 154 struct btrfs_path *path; 155 u8 *csum; 156 u64 offset = 0; 157 u64 item_start_offset = 0; 158 u64 item_last_offset = 0; 159 u64 disk_bytenr; 160 u64 page_bytes_left; 161 u32 diff; 162 int nblocks; 163 int count = 0; 164 u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); 165 166 path = btrfs_alloc_path(); 167 if (!path) 168 return BLK_STS_RESOURCE; 169 170 nblocks = bio->bi_iter.bi_size >> inode->i_sb->s_blocksize_bits; 171 if (!dst) { 172 if (nblocks * csum_size > BTRFS_BIO_INLINE_CSUM_SIZE) { 173 btrfs_bio->csum = kmalloc_array(nblocks, csum_size, 174 GFP_NOFS); 175 if (!btrfs_bio->csum) { 176 btrfs_free_path(path); 177 return BLK_STS_RESOURCE; 178 } 179 } else { 180 btrfs_bio->csum = btrfs_bio->csum_inline; 181 } 182 csum = btrfs_bio->csum; 183 } else { 184 csum = (u8 *)dst; 185 } 186 187 if (bio->bi_iter.bi_size > PAGE_SIZE * 8) 188 path->reada = READA_FORWARD; 189 190 /* 191 * the free space stuff is only read when it hasn't been 192 * updated in the current transaction. So, we can safely 193 * read from the commit root and sidestep a nasty deadlock 194 * between reading the free space cache and updating the csum tree. 195 */ 196 if (btrfs_is_free_space_inode(BTRFS_I(inode))) { 197 path->search_commit_root = 1; 198 path->skip_locking = 1; 199 } 200 201 disk_bytenr = (u64)bio->bi_iter.bi_sector << 9; 202 if (dio) 203 offset = logical_offset; 204 205 bio_for_each_segment(bvec, bio, iter) { 206 page_bytes_left = bvec.bv_len; 207 if (count) 208 goto next; 209 210 if (!dio) 211 offset = page_offset(bvec.bv_page) + bvec.bv_offset; 212 count = btrfs_find_ordered_sum(inode, offset, disk_bytenr, 213 (u32 *)csum, nblocks); 214 if (count) 215 goto found; 216 217 if (!item || disk_bytenr < item_start_offset || 218 disk_bytenr >= item_last_offset) { 219 struct btrfs_key found_key; 220 u32 item_size; 221 222 if (item) 223 btrfs_release_path(path); 224 item = btrfs_lookup_csum(NULL, fs_info->csum_root, 225 path, disk_bytenr, 0); 226 if (IS_ERR(item)) { 227 count = 1; 228 memset(csum, 0, csum_size); 229 if (BTRFS_I(inode)->root->root_key.objectid == 230 BTRFS_DATA_RELOC_TREE_OBJECTID) { 231 set_extent_bits(io_tree, offset, 232 offset + fs_info->sectorsize - 1, 233 EXTENT_NODATASUM); 234 } else { 235 btrfs_info_rl(fs_info, 236 "no csum found for inode %llu start %llu", 237 btrfs_ino(BTRFS_I(inode)), offset); 238 } 239 item = NULL; 240 btrfs_release_path(path); 241 goto found; 242 } 243 btrfs_item_key_to_cpu(path->nodes[0], &found_key, 244 path->slots[0]); 245 246 item_start_offset = found_key.offset; 247 item_size = btrfs_item_size_nr(path->nodes[0], 248 path->slots[0]); 249 item_last_offset = item_start_offset + 250 (item_size / csum_size) * 251 fs_info->sectorsize; 252 item = btrfs_item_ptr(path->nodes[0], path->slots[0], 253 struct btrfs_csum_item); 254 } 255 /* 256 * this byte range must be able to fit inside 257 * a single leaf so it will also fit inside a u32 258 */ 259 diff = disk_bytenr - item_start_offset; 260 diff = diff / fs_info->sectorsize; 261 diff = diff * csum_size; 262 count = min_t(int, nblocks, (item_last_offset - disk_bytenr) >> 263 inode->i_sb->s_blocksize_bits); 264 read_extent_buffer(path->nodes[0], csum, 265 ((unsigned long)item) + diff, 266 csum_size * count); 267 found: 268 csum += count * csum_size; 269 nblocks -= count; 270 next: 271 while (count--) { 272 disk_bytenr += fs_info->sectorsize; 273 offset += fs_info->sectorsize; 274 page_bytes_left -= fs_info->sectorsize; 275 if (!page_bytes_left) 276 break; /* move to next bio */ 277 } 278 } 279 280 WARN_ON_ONCE(count); 281 btrfs_free_path(path); 282 return 0; 283 } 284 285 blk_status_t btrfs_lookup_bio_sums(struct inode *inode, struct bio *bio, u32 *dst) 286 { 287 return __btrfs_lookup_bio_sums(inode, bio, 0, dst, 0); 288 } 289 290 blk_status_t btrfs_lookup_bio_sums_dio(struct inode *inode, struct bio *bio, u64 offset) 291 { 292 return __btrfs_lookup_bio_sums(inode, bio, offset, NULL, 1); 293 } 294 295 int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end, 296 struct list_head *list, int search_commit) 297 { 298 struct btrfs_fs_info *fs_info = root->fs_info; 299 struct btrfs_key key; 300 struct btrfs_path *path; 301 struct extent_buffer *leaf; 302 struct btrfs_ordered_sum *sums; 303 struct btrfs_csum_item *item; 304 LIST_HEAD(tmplist); 305 unsigned long offset; 306 int ret; 307 size_t size; 308 u64 csum_end; 309 u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); 310 311 ASSERT(IS_ALIGNED(start, fs_info->sectorsize) && 312 IS_ALIGNED(end + 1, fs_info->sectorsize)); 313 314 path = btrfs_alloc_path(); 315 if (!path) 316 return -ENOMEM; 317 318 if (search_commit) { 319 path->skip_locking = 1; 320 path->reada = READA_FORWARD; 321 path->search_commit_root = 1; 322 } 323 324 key.objectid = BTRFS_EXTENT_CSUM_OBJECTID; 325 key.offset = start; 326 key.type = BTRFS_EXTENT_CSUM_KEY; 327 328 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); 329 if (ret < 0) 330 goto fail; 331 if (ret > 0 && path->slots[0] > 0) { 332 leaf = path->nodes[0]; 333 btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1); 334 if (key.objectid == BTRFS_EXTENT_CSUM_OBJECTID && 335 key.type == BTRFS_EXTENT_CSUM_KEY) { 336 offset = (start - key.offset) >> 337 fs_info->sb->s_blocksize_bits; 338 if (offset * csum_size < 339 btrfs_item_size_nr(leaf, path->slots[0] - 1)) 340 path->slots[0]--; 341 } 342 } 343 344 while (start <= end) { 345 leaf = path->nodes[0]; 346 if (path->slots[0] >= btrfs_header_nritems(leaf)) { 347 ret = btrfs_next_leaf(root, path); 348 if (ret < 0) 349 goto fail; 350 if (ret > 0) 351 break; 352 leaf = path->nodes[0]; 353 } 354 355 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); 356 if (key.objectid != BTRFS_EXTENT_CSUM_OBJECTID || 357 key.type != BTRFS_EXTENT_CSUM_KEY || 358 key.offset > end) 359 break; 360 361 if (key.offset > start) 362 start = key.offset; 363 364 size = btrfs_item_size_nr(leaf, path->slots[0]); 365 csum_end = key.offset + (size / csum_size) * fs_info->sectorsize; 366 if (csum_end <= start) { 367 path->slots[0]++; 368 continue; 369 } 370 371 csum_end = min(csum_end, end + 1); 372 item = btrfs_item_ptr(path->nodes[0], path->slots[0], 373 struct btrfs_csum_item); 374 while (start < csum_end) { 375 size = min_t(size_t, csum_end - start, 376 MAX_ORDERED_SUM_BYTES(fs_info)); 377 sums = kzalloc(btrfs_ordered_sum_size(fs_info, size), 378 GFP_NOFS); 379 if (!sums) { 380 ret = -ENOMEM; 381 goto fail; 382 } 383 384 sums->bytenr = start; 385 sums->len = (int)size; 386 387 offset = (start - key.offset) >> 388 fs_info->sb->s_blocksize_bits; 389 offset *= csum_size; 390 size >>= fs_info->sb->s_blocksize_bits; 391 392 read_extent_buffer(path->nodes[0], 393 sums->sums, 394 ((unsigned long)item) + offset, 395 csum_size * size); 396 397 start += fs_info->sectorsize * size; 398 list_add_tail(&sums->list, &tmplist); 399 } 400 path->slots[0]++; 401 } 402 ret = 0; 403 fail: 404 while (ret < 0 && !list_empty(&tmplist)) { 405 sums = list_entry(tmplist.next, struct btrfs_ordered_sum, list); 406 list_del(&sums->list); 407 kfree(sums); 408 } 409 list_splice_tail(&tmplist, list); 410 411 btrfs_free_path(path); 412 return ret; 413 } 414 415 blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio, 416 u64 file_start, int contig) 417 { 418 struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); 419 struct btrfs_ordered_sum *sums; 420 struct btrfs_ordered_extent *ordered = NULL; 421 char *data; 422 struct bvec_iter iter; 423 struct bio_vec bvec; 424 int index; 425 int nr_sectors; 426 unsigned long total_bytes = 0; 427 unsigned long this_sum_bytes = 0; 428 int i; 429 u64 offset; 430 431 sums = kzalloc(btrfs_ordered_sum_size(fs_info, bio->bi_iter.bi_size), 432 GFP_NOFS); 433 if (!sums) 434 return BLK_STS_RESOURCE; 435 436 sums->len = bio->bi_iter.bi_size; 437 INIT_LIST_HEAD(&sums->list); 438 439 if (contig) 440 offset = file_start; 441 else 442 offset = 0; /* shut up gcc */ 443 444 sums->bytenr = (u64)bio->bi_iter.bi_sector << 9; 445 index = 0; 446 447 bio_for_each_segment(bvec, bio, iter) { 448 if (!contig) 449 offset = page_offset(bvec.bv_page) + bvec.bv_offset; 450 451 if (!ordered) { 452 ordered = btrfs_lookup_ordered_extent(inode, offset); 453 BUG_ON(!ordered); /* Logic error */ 454 } 455 456 data = kmap_atomic(bvec.bv_page); 457 458 nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info, 459 bvec.bv_len + fs_info->sectorsize 460 - 1); 461 462 for (i = 0; i < nr_sectors; i++) { 463 if (offset >= ordered->file_offset + ordered->len || 464 offset < ordered->file_offset) { 465 unsigned long bytes_left; 466 467 kunmap_atomic(data); 468 sums->len = this_sum_bytes; 469 this_sum_bytes = 0; 470 btrfs_add_ordered_sum(inode, ordered, sums); 471 btrfs_put_ordered_extent(ordered); 472 473 bytes_left = bio->bi_iter.bi_size - total_bytes; 474 475 sums = kzalloc(btrfs_ordered_sum_size(fs_info, bytes_left), 476 GFP_NOFS); 477 BUG_ON(!sums); /* -ENOMEM */ 478 sums->len = bytes_left; 479 ordered = btrfs_lookup_ordered_extent(inode, 480 offset); 481 ASSERT(ordered); /* Logic error */ 482 sums->bytenr = ((u64)bio->bi_iter.bi_sector << 9) 483 + total_bytes; 484 index = 0; 485 486 data = kmap_atomic(bvec.bv_page); 487 } 488 489 sums->sums[index] = ~(u32)0; 490 sums->sums[index] 491 = btrfs_csum_data(data + bvec.bv_offset 492 + (i * fs_info->sectorsize), 493 sums->sums[index], 494 fs_info->sectorsize); 495 btrfs_csum_final(sums->sums[index], 496 (char *)(sums->sums + index)); 497 index++; 498 offset += fs_info->sectorsize; 499 this_sum_bytes += fs_info->sectorsize; 500 total_bytes += fs_info->sectorsize; 501 } 502 503 kunmap_atomic(data); 504 } 505 this_sum_bytes = 0; 506 btrfs_add_ordered_sum(inode, ordered, sums); 507 btrfs_put_ordered_extent(ordered); 508 return 0; 509 } 510 511 /* 512 * helper function for csum removal, this expects the 513 * key to describe the csum pointed to by the path, and it expects 514 * the csum to overlap the range [bytenr, len] 515 * 516 * The csum should not be entirely contained in the range and the 517 * range should not be entirely contained in the csum. 518 * 519 * This calls btrfs_truncate_item with the correct args based on the 520 * overlap, and fixes up the key as required. 521 */ 522 static noinline void truncate_one_csum(struct btrfs_fs_info *fs_info, 523 struct btrfs_path *path, 524 struct btrfs_key *key, 525 u64 bytenr, u64 len) 526 { 527 struct extent_buffer *leaf; 528 u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); 529 u64 csum_end; 530 u64 end_byte = bytenr + len; 531 u32 blocksize_bits = fs_info->sb->s_blocksize_bits; 532 533 leaf = path->nodes[0]; 534 csum_end = btrfs_item_size_nr(leaf, path->slots[0]) / csum_size; 535 csum_end <<= fs_info->sb->s_blocksize_bits; 536 csum_end += key->offset; 537 538 if (key->offset < bytenr && csum_end <= end_byte) { 539 /* 540 * [ bytenr - len ] 541 * [ ] 542 * [csum ] 543 * A simple truncate off the end of the item 544 */ 545 u32 new_size = (bytenr - key->offset) >> blocksize_bits; 546 new_size *= csum_size; 547 btrfs_truncate_item(fs_info, path, new_size, 1); 548 } else if (key->offset >= bytenr && csum_end > end_byte && 549 end_byte > key->offset) { 550 /* 551 * [ bytenr - len ] 552 * [ ] 553 * [csum ] 554 * we need to truncate from the beginning of the csum 555 */ 556 u32 new_size = (csum_end - end_byte) >> blocksize_bits; 557 new_size *= csum_size; 558 559 btrfs_truncate_item(fs_info, path, new_size, 0); 560 561 key->offset = end_byte; 562 btrfs_set_item_key_safe(fs_info, path, key); 563 } else { 564 BUG(); 565 } 566 } 567 568 /* 569 * deletes the csum items from the csum tree for a given 570 * range of bytes. 571 */ 572 int btrfs_del_csums(struct btrfs_trans_handle *trans, 573 struct btrfs_fs_info *fs_info, u64 bytenr, u64 len) 574 { 575 struct btrfs_root *root = fs_info->csum_root; 576 struct btrfs_path *path; 577 struct btrfs_key key; 578 u64 end_byte = bytenr + len; 579 u64 csum_end; 580 struct extent_buffer *leaf; 581 int ret; 582 u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); 583 int blocksize_bits = fs_info->sb->s_blocksize_bits; 584 585 path = btrfs_alloc_path(); 586 if (!path) 587 return -ENOMEM; 588 589 while (1) { 590 key.objectid = BTRFS_EXTENT_CSUM_OBJECTID; 591 key.offset = end_byte - 1; 592 key.type = BTRFS_EXTENT_CSUM_KEY; 593 594 path->leave_spinning = 1; 595 ret = btrfs_search_slot(trans, root, &key, path, -1, 1); 596 if (ret > 0) { 597 if (path->slots[0] == 0) 598 break; 599 path->slots[0]--; 600 } else if (ret < 0) { 601 break; 602 } 603 604 leaf = path->nodes[0]; 605 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); 606 607 if (key.objectid != BTRFS_EXTENT_CSUM_OBJECTID || 608 key.type != BTRFS_EXTENT_CSUM_KEY) { 609 break; 610 } 611 612 if (key.offset >= end_byte) 613 break; 614 615 csum_end = btrfs_item_size_nr(leaf, path->slots[0]) / csum_size; 616 csum_end <<= blocksize_bits; 617 csum_end += key.offset; 618 619 /* this csum ends before we start, we're done */ 620 if (csum_end <= bytenr) 621 break; 622 623 /* delete the entire item, it is inside our range */ 624 if (key.offset >= bytenr && csum_end <= end_byte) { 625 int del_nr = 1; 626 627 /* 628 * Check how many csum items preceding this one in this 629 * leaf correspond to our range and then delete them all 630 * at once. 631 */ 632 if (key.offset > bytenr && path->slots[0] > 0) { 633 int slot = path->slots[0] - 1; 634 635 while (slot >= 0) { 636 struct btrfs_key pk; 637 638 btrfs_item_key_to_cpu(leaf, &pk, slot); 639 if (pk.offset < bytenr || 640 pk.type != BTRFS_EXTENT_CSUM_KEY || 641 pk.objectid != 642 BTRFS_EXTENT_CSUM_OBJECTID) 643 break; 644 path->slots[0] = slot; 645 del_nr++; 646 key.offset = pk.offset; 647 slot--; 648 } 649 } 650 ret = btrfs_del_items(trans, root, path, 651 path->slots[0], del_nr); 652 if (ret) 653 goto out; 654 if (key.offset == bytenr) 655 break; 656 } else if (key.offset < bytenr && csum_end > end_byte) { 657 unsigned long offset; 658 unsigned long shift_len; 659 unsigned long item_offset; 660 /* 661 * [ bytenr - len ] 662 * [csum ] 663 * 664 * Our bytes are in the middle of the csum, 665 * we need to split this item and insert a new one. 666 * 667 * But we can't drop the path because the 668 * csum could change, get removed, extended etc. 669 * 670 * The trick here is the max size of a csum item leaves 671 * enough room in the tree block for a single 672 * item header. So, we split the item in place, 673 * adding a new header pointing to the existing 674 * bytes. Then we loop around again and we have 675 * a nicely formed csum item that we can neatly 676 * truncate. 677 */ 678 offset = (bytenr - key.offset) >> blocksize_bits; 679 offset *= csum_size; 680 681 shift_len = (len >> blocksize_bits) * csum_size; 682 683 item_offset = btrfs_item_ptr_offset(leaf, 684 path->slots[0]); 685 686 memzero_extent_buffer(leaf, item_offset + offset, 687 shift_len); 688 key.offset = bytenr; 689 690 /* 691 * btrfs_split_item returns -EAGAIN when the 692 * item changed size or key 693 */ 694 ret = btrfs_split_item(trans, root, path, &key, offset); 695 if (ret && ret != -EAGAIN) { 696 btrfs_abort_transaction(trans, ret); 697 goto out; 698 } 699 700 key.offset = end_byte - 1; 701 } else { 702 truncate_one_csum(fs_info, path, &key, bytenr, len); 703 if (key.offset < bytenr) 704 break; 705 } 706 btrfs_release_path(path); 707 } 708 ret = 0; 709 out: 710 btrfs_free_path(path); 711 return ret; 712 } 713 714 int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans, 715 struct btrfs_root *root, 716 struct btrfs_ordered_sum *sums) 717 { 718 struct btrfs_fs_info *fs_info = root->fs_info; 719 struct btrfs_key file_key; 720 struct btrfs_key found_key; 721 struct btrfs_path *path; 722 struct btrfs_csum_item *item; 723 struct btrfs_csum_item *item_end; 724 struct extent_buffer *leaf = NULL; 725 u64 next_offset; 726 u64 total_bytes = 0; 727 u64 csum_offset; 728 u64 bytenr; 729 u32 nritems; 730 u32 ins_size; 731 int index = 0; 732 int found_next; 733 int ret; 734 u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); 735 736 path = btrfs_alloc_path(); 737 if (!path) 738 return -ENOMEM; 739 again: 740 next_offset = (u64)-1; 741 found_next = 0; 742 bytenr = sums->bytenr + total_bytes; 743 file_key.objectid = BTRFS_EXTENT_CSUM_OBJECTID; 744 file_key.offset = bytenr; 745 file_key.type = BTRFS_EXTENT_CSUM_KEY; 746 747 item = btrfs_lookup_csum(trans, root, path, bytenr, 1); 748 if (!IS_ERR(item)) { 749 ret = 0; 750 leaf = path->nodes[0]; 751 item_end = btrfs_item_ptr(leaf, path->slots[0], 752 struct btrfs_csum_item); 753 item_end = (struct btrfs_csum_item *)((char *)item_end + 754 btrfs_item_size_nr(leaf, path->slots[0])); 755 goto found; 756 } 757 ret = PTR_ERR(item); 758 if (ret != -EFBIG && ret != -ENOENT) 759 goto fail_unlock; 760 761 if (ret == -EFBIG) { 762 u32 item_size; 763 /* we found one, but it isn't big enough yet */ 764 leaf = path->nodes[0]; 765 item_size = btrfs_item_size_nr(leaf, path->slots[0]); 766 if ((item_size / csum_size) >= 767 MAX_CSUM_ITEMS(fs_info, csum_size)) { 768 /* already at max size, make a new one */ 769 goto insert; 770 } 771 } else { 772 int slot = path->slots[0] + 1; 773 /* we didn't find a csum item, insert one */ 774 nritems = btrfs_header_nritems(path->nodes[0]); 775 if (!nritems || (path->slots[0] >= nritems - 1)) { 776 ret = btrfs_next_leaf(root, path); 777 if (ret == 1) 778 found_next = 1; 779 if (ret != 0) 780 goto insert; 781 slot = path->slots[0]; 782 } 783 btrfs_item_key_to_cpu(path->nodes[0], &found_key, slot); 784 if (found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID || 785 found_key.type != BTRFS_EXTENT_CSUM_KEY) { 786 found_next = 1; 787 goto insert; 788 } 789 next_offset = found_key.offset; 790 found_next = 1; 791 goto insert; 792 } 793 794 /* 795 * at this point, we know the tree has an item, but it isn't big 796 * enough yet to put our csum in. Grow it 797 */ 798 btrfs_release_path(path); 799 ret = btrfs_search_slot(trans, root, &file_key, path, 800 csum_size, 1); 801 if (ret < 0) 802 goto fail_unlock; 803 804 if (ret > 0) { 805 if (path->slots[0] == 0) 806 goto insert; 807 path->slots[0]--; 808 } 809 810 leaf = path->nodes[0]; 811 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); 812 csum_offset = (bytenr - found_key.offset) >> 813 fs_info->sb->s_blocksize_bits; 814 815 if (found_key.type != BTRFS_EXTENT_CSUM_KEY || 816 found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID || 817 csum_offset >= MAX_CSUM_ITEMS(fs_info, csum_size)) { 818 goto insert; 819 } 820 821 if (csum_offset == btrfs_item_size_nr(leaf, path->slots[0]) / 822 csum_size) { 823 int extend_nr; 824 u64 tmp; 825 u32 diff; 826 u32 free_space; 827 828 if (btrfs_leaf_free_space(fs_info, leaf) < 829 sizeof(struct btrfs_item) + csum_size * 2) 830 goto insert; 831 832 free_space = btrfs_leaf_free_space(fs_info, leaf) - 833 sizeof(struct btrfs_item) - csum_size; 834 tmp = sums->len - total_bytes; 835 tmp >>= fs_info->sb->s_blocksize_bits; 836 WARN_ON(tmp < 1); 837 838 extend_nr = max_t(int, 1, (int)tmp); 839 diff = (csum_offset + extend_nr) * csum_size; 840 diff = min(diff, 841 MAX_CSUM_ITEMS(fs_info, csum_size) * csum_size); 842 843 diff = diff - btrfs_item_size_nr(leaf, path->slots[0]); 844 diff = min(free_space, diff); 845 diff /= csum_size; 846 diff *= csum_size; 847 848 btrfs_extend_item(fs_info, path, diff); 849 ret = 0; 850 goto csum; 851 } 852 853 insert: 854 btrfs_release_path(path); 855 csum_offset = 0; 856 if (found_next) { 857 u64 tmp; 858 859 tmp = sums->len - total_bytes; 860 tmp >>= fs_info->sb->s_blocksize_bits; 861 tmp = min(tmp, (next_offset - file_key.offset) >> 862 fs_info->sb->s_blocksize_bits); 863 864 tmp = max_t(u64, 1, tmp); 865 tmp = min_t(u64, tmp, MAX_CSUM_ITEMS(fs_info, csum_size)); 866 ins_size = csum_size * tmp; 867 } else { 868 ins_size = csum_size; 869 } 870 path->leave_spinning = 1; 871 ret = btrfs_insert_empty_item(trans, root, path, &file_key, 872 ins_size); 873 path->leave_spinning = 0; 874 if (ret < 0) 875 goto fail_unlock; 876 if (WARN_ON(ret != 0)) 877 goto fail_unlock; 878 leaf = path->nodes[0]; 879 csum: 880 item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item); 881 item_end = (struct btrfs_csum_item *)((unsigned char *)item + 882 btrfs_item_size_nr(leaf, path->slots[0])); 883 item = (struct btrfs_csum_item *)((unsigned char *)item + 884 csum_offset * csum_size); 885 found: 886 ins_size = (u32)(sums->len - total_bytes) >> 887 fs_info->sb->s_blocksize_bits; 888 ins_size *= csum_size; 889 ins_size = min_t(u32, (unsigned long)item_end - (unsigned long)item, 890 ins_size); 891 write_extent_buffer(leaf, sums->sums + index, (unsigned long)item, 892 ins_size); 893 894 ins_size /= csum_size; 895 total_bytes += ins_size * fs_info->sectorsize; 896 index += ins_size; 897 898 btrfs_mark_buffer_dirty(path->nodes[0]); 899 if (total_bytes < sums->len) { 900 btrfs_release_path(path); 901 cond_resched(); 902 goto again; 903 } 904 out: 905 btrfs_free_path(path); 906 return ret; 907 908 fail_unlock: 909 goto out; 910 } 911 912 void btrfs_extent_item_to_extent_map(struct btrfs_inode *inode, 913 const struct btrfs_path *path, 914 struct btrfs_file_extent_item *fi, 915 const bool new_inline, 916 struct extent_map *em) 917 { 918 struct btrfs_fs_info *fs_info = inode->root->fs_info; 919 struct btrfs_root *root = inode->root; 920 struct extent_buffer *leaf = path->nodes[0]; 921 const int slot = path->slots[0]; 922 struct btrfs_key key; 923 u64 extent_start, extent_end; 924 u64 bytenr; 925 u8 type = btrfs_file_extent_type(leaf, fi); 926 int compress_type = btrfs_file_extent_compression(leaf, fi); 927 928 em->bdev = fs_info->fs_devices->latest_bdev; 929 btrfs_item_key_to_cpu(leaf, &key, slot); 930 extent_start = key.offset; 931 932 if (type == BTRFS_FILE_EXTENT_REG || 933 type == BTRFS_FILE_EXTENT_PREALLOC) { 934 extent_end = extent_start + 935 btrfs_file_extent_num_bytes(leaf, fi); 936 } else if (type == BTRFS_FILE_EXTENT_INLINE) { 937 size_t size; 938 size = btrfs_file_extent_ram_bytes(leaf, fi); 939 extent_end = ALIGN(extent_start + size, 940 fs_info->sectorsize); 941 } 942 943 em->ram_bytes = btrfs_file_extent_ram_bytes(leaf, fi); 944 if (type == BTRFS_FILE_EXTENT_REG || 945 type == BTRFS_FILE_EXTENT_PREALLOC) { 946 em->start = extent_start; 947 em->len = extent_end - extent_start; 948 em->orig_start = extent_start - 949 btrfs_file_extent_offset(leaf, fi); 950 em->orig_block_len = btrfs_file_extent_disk_num_bytes(leaf, fi); 951 bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); 952 if (bytenr == 0) { 953 em->block_start = EXTENT_MAP_HOLE; 954 return; 955 } 956 if (compress_type != BTRFS_COMPRESS_NONE) { 957 set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); 958 em->compress_type = compress_type; 959 em->block_start = bytenr; 960 em->block_len = em->orig_block_len; 961 } else { 962 bytenr += btrfs_file_extent_offset(leaf, fi); 963 em->block_start = bytenr; 964 em->block_len = em->len; 965 if (type == BTRFS_FILE_EXTENT_PREALLOC) 966 set_bit(EXTENT_FLAG_PREALLOC, &em->flags); 967 } 968 } else if (type == BTRFS_FILE_EXTENT_INLINE) { 969 em->block_start = EXTENT_MAP_INLINE; 970 em->start = extent_start; 971 em->len = extent_end - extent_start; 972 /* 973 * Initialize orig_start and block_len with the same values 974 * as in inode.c:btrfs_get_extent(). 975 */ 976 em->orig_start = EXTENT_MAP_HOLE; 977 em->block_len = (u64)-1; 978 if (!new_inline && compress_type != BTRFS_COMPRESS_NONE) { 979 set_bit(EXTENT_FLAG_COMPRESSED, &em->flags); 980 em->compress_type = compress_type; 981 } 982 } else { 983 btrfs_err(fs_info, 984 "unknown file extent item type %d, inode %llu, offset %llu, " 985 "root %llu", type, btrfs_ino(inode), extent_start, 986 root->root_key.objectid); 987 } 988 } 989